Work item:
Y.IIoT-AIA
Subject/title:
Requirements and framework for AI agent system in industrial IoT environment
Status:
Under study
Approval process:
AAP
Type of work item:
Recommendation
Version:
New
Equivalent number:
-
Timing:
2027-Q4 (Medium priority)
Liaison:
ITU-T SG21, IEEE C/AISC , ISO/IEC JTC 1/SC 42, ETSI TC SAI and CEN/CLC/JTC 21
Supporting members:
China Telecom, Zhejiang Dahua Technology, Shenzhen SmartMore Technology Co., Ltd, ETRI
Summary:
Industries environment is a specific scene with multiple sensors, communication network and applications, typical industry applications include defect detection, industrial design, business training, industrial robot navigation, production process simulation, environment simulation and layout optimization. A system with AI agents is helpful for reducing labour-intensive tasks and providing professional guidance for technical experts.
IoT sensors and cameras captures large amount of data. However, the industrial applications cannot meet the practical business need because of the following pain points: (a) poor data quality (b) poor AI model interoperability (c) poor management of API service resources (d) poor server service discovery.
AI agent system is a type of system that can autonomously perceive the environment, reason and implement actions. They have the comprehensive ability to obtain environment input, make judgements and decisions and influence environmental outputs. AI agent can incorporate various technologies including foundation models, Retrieval Augmented Generation (RAG) technology, Chaine of Thought (CoT) technology, workflow automation integration and knowledge-based invocation, to serve industrial enterprises applications, including defect detection, industrial design, business training, industrial robot navigation, production process simulation, environment simulation and layout optimization.
By defining the functional reference framework and requirements of AI agent system, the enterprise can flexibly incorporate internal and external data resources, APIs, foundation models and to promote the mutual understanding and strengthen confidence of AI agent system.
Comment:
-
Historic references: